Benchmarking genomic variant calling tools in inbred mouse strains: recommendations and considerations

With the growing affordability of whole-genome sequencing, variant identification has become an increasingly common task in modern genomics. In recent years, the number of software packages available for variant calling has rapidly increased. Understanding the benefits and drawbacks of different tools is important in setting leading practices. These considerations are especially crucial in model organism research, as many variant calling tools assume outbred genomes and implicit heterozygosity, conditions that do not apply to inbred laboratory models. Here, we present an analysis of multiple widely used variant calling tools and their performance in the simulated genomes of the C57BL/6J inbred laboratory mouse and 9 non-reference strains. Our findings reveal a tradeoff between the recall and precision of different tools. Balancing these considerations, we show that an optimal variant call set is obtained by using an ensemble approach focused on the intersection of variants reported by multiple callers. However, specific variant calling recommendations vary by strain and analytical goals. Further, we identify empirical filters for improving the performance of different variant calling tools, both for the discovery of rare variants and in the identification of strain polymorphisms. Overall, our simulation-based analysis provides best practices for calling and filtering genomic variants in inbred organisms, particularly laboratory mice.